CMIP6与CMIP5对历史大气层顶和地表辐射收支模拟的时空对比

基于云和地球辐射能量系统观测数据集（CERES),对比分析了耦合模式比较计划第五（CMIP5）和第六阶段（CMIP6）模拟的历史大气层顶和地表辐射收支的年际变化和空间分布,明确了多模式间不确定性大的关键区域。结果表明：在年际尺度上,除地表向上长波辐射外,CMIP6的辐射分量的集合均值较CMIP5更接近于CERES观测值,全球地表向下短波辐射的高估和大气逆辐射的低估在CMIP6中分别降低了1.9 W/m²和3.3 W/m²。除大气逆辐射外,CMIP6的辐射分量在多模式间的一致性较CMIP5提高。在北极,CMIP6对大气层顶反射短波、大气层顶出射长波和地表向下短波辐射的模拟偏差较CMIP5大。在南北纬60°,CMIP6对大气逆辐射的模拟偏差较CMIP5大。其他区域CMIP6的辐射分量更接近CERES观测值。CMIP6模拟的地表向下短波辐射和大气逆辐射的不确定性较大区域面积较CMIP5减小,但不确定性极大区域面积无变化。地表净辐射的不确定性空间分布在两代CMIP间变化甚小。青藏高原、赤道太平洋、热带雨林、阿拉伯半岛和南极洲沿海依然是地球系统模式模拟辐射收支不确定性极大的关键区域。

Comparison between CMIP6 and CMIP5 models in simulating historical spatiotemporal variations in radiation budgets at the top of atmosphere and the surface

Based on Clouds and the Earth’s Radiant Energy System (CERES) Energy Balanced and Filled (EBAF) Dataset Edition 4.1, historical spatiotemporal variations in radiation budgets at the top of atmosphere (TOA) and the Earth surface were compared between Coupled Model Intercomparison Project phase 5 (CMIP5) and phase 6 (CMIP6). Regions with high inter-model variability were identified in two CMIPs. The results show that ensemble means of radiation components, except surface upward longwave radiation in CMIP6 are in better agreement with CERES EBAF 4.1. Except downward longwave radiation at the surface, lower inter-model spread for other radiation components are found in CMIP6. Overestimation in global mean surface downward shortwave is reduced by 1.9 W/m², and underestimation in global mean downward longwave radiation is reduced by 3.3 W/m², in CMIP6. Spatially, larger deviations are found in TOA reflected shortwave and outgoing longwave radiation, as well as surface downward shortwave radiation around the North Pole in CMIP6 compared to CMIP5. Worse simulation is also found in CMIP6 for surface downward longwave radiation around 60° latitudes. In other regions, CMIP6 radiation components agree better with CERES EBAF 4.1 than CMIP5. Regions with relative large inter-model variability for surface downward shortwave and downward longwave radiation shrink from CMIP5 to CMIP6. However, regions with extreme large inter-model variability still keep almost the same for the two components in two CMIPs. Two CMIPs are similar in spatial distribution with large inter-model variability for surface net radiation. Tibet Plateau, equatorial Pacific, tropical rainforest, Arabian Peninsula and Antarctic coasts are important regions with extreme large inter-model variability for simulating radiation budgets at the TOA and the surface by Earth System Models.